CN106126746A - High-quality node detecting method and system in a kind of social networks - Google Patents
High-quality node detecting method and system in a kind of social networks Download PDFInfo
- Publication number
- CN106126746A CN106126746A CN201610550702.6A CN201610550702A CN106126746A CN 106126746 A CN106126746 A CN 106126746A CN 201610550702 A CN201610550702 A CN 201610550702A CN 106126746 A CN106126746 A CN 106126746A
- Authority
- CN
- China
- Prior art keywords
- node
- quality
- degree
- social networks
- detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000001514 detection method Methods 0.000 claims abstract description 101
- 238000013507 mapping Methods 0.000 claims abstract description 22
- 230000008569 process Effects 0.000 claims abstract description 22
- 238000012549 training Methods 0.000 claims abstract description 13
- 239000000284 extract Substances 0.000 claims abstract description 10
- 238000002474 experimental method Methods 0.000 claims abstract description 7
- 238000012512 characterization method Methods 0.000 claims abstract description 6
- 239000011159 matrix material Substances 0.000 claims description 26
- 230000002452 interceptive effect Effects 0.000 claims description 17
- 239000000523 sample Substances 0.000 claims description 16
- 238000000605 extraction Methods 0.000 claims description 13
- 238000011160 research Methods 0.000 description 8
- 230000006399 behavior Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 239000000047 product Substances 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000009191 jumping Effects 0.000 description 3
- 208000020016 psychiatric disease Diseases 0.000 description 3
- 230000007480 spreading Effects 0.000 description 3
- 238000003892 spreading Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 244000144992 flock Species 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000006698 induction Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000006855 networking Effects 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 235000012364 Peperomia pellucida Nutrition 0.000 description 1
- 240000007711 Peperomia pellucida Species 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 238000009415 formwork Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000005304 joining Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000002040 relaxant effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 239000002356 single layer Substances 0.000 description 1
- 230000003997 social interaction Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 230000003442 weekly effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Signal Processing (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Economics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
High-quality node detecting method in a kind of social networks, it comprises the steps: S1, extracts the social networks node set at the high-quality node place needing detection;S2, the social networks node in social networks node set is set up the node mapping relations of social networks;S3, extract detection high-quality node diagnostic according to the mechanics of high-quality node;The characterization rules model of high-quality node detection is set up according to the high-quality node diagnostic extracted;S4, social networks node is grouped as experiment sample, then carries out station work and node-classification;S5, the result detecting high-quality node are estimated and feedback result, and the rule not meeting detection high-quality node diagnostic are corrected during training repeatedly, thus reach the purpose being optimized model;S6, will optimize after model turn again to step S4 until the detection progress of high-quality node exceedes setting threshold value thus completes high-quality node detection process in whole social networks.
Description
Technical field
The present invention relates to key node detection study technical field in social networks, particularly to excellent in a kind of social networks
Matter node detecting method and system.
Background technology
In recent years, owing to the research of social networks is the most popular, based on the detection in social network-i i-platform with find excellent
The research of matter customer problem obtains the attention of people the most day by day.By the excavation to this kind of user, can be with these high-qualitys user
Set up and more directly associate the social value contained with acquisition.Such as, teacher hankers after finding that outstanding student carries out emphasis training
Supporting, businessman hankers after finding that the affiliate of high-quality carries out business associate, and financial industry is hankered after finding that the client of high-quality is with reality
Existing economic worth, working HR for many years is good at finding that the outstanding talent comes for enterprises service etc..So, possess quick as one
The observer of sharp eyes light, they be again by which kind of knowledge and experience go to find top-tier customer in the industry?These
Problem key problem the most to be solved by this invention: by extract sharp-sighted observer knowledge that high-quality be the user discover that and
Based on experience, the feature that therefrom extraction high-quality user is possessed sets up correlation rule and feature identification storehouse.To observer found that
The problem of high-quality user is incorporated in the research field of social networks, excavates as theoretical foundation with the figure of social networks, this is asked
Topic is converted in social networks the problem finding and detecting high-quality node, thus defines and a set of solved this challenge
Method system certainly.
By the reading of domestic and international list of references and the analysis of present Research are found, build with social network diagram Extracting Knowledge
Formwork erection type, actually rare to find the research papers of high-quality node, the list of references being closely related with the present invention is just more difficult to
To seek.Therefore, problem to be solved by this invention, not only have certain in the domestic and international present Research involved by this problem
Novelty, and solving in the method for problem, approach and thinking, having more its originality.By dabbling and phase of the present invention
The list of references closed, mainly studying a question and studying of relating to a little can be summarized as follows.Wherein, optimum message issuing time:
Nemanja Spasojevic et al. has cooked up a kind of problem being called when-to-post, and the target solving such problem is
The Best Times that in social networks to be found out, user gives out information, thus it is worth the feedback probability of audience to reach to maximize.In order to
Allowing reader it will be appreciated that this complex nature of the problem, they have investigated the change of user behavior in terms of the response time that gives out information
Situation, and compare the user in different cities across a network and trans-city reaction row on Twitter and Facebook weekly
For.This analytical mathematics is given out information with 1,000,000,000 and carries out implementing checking by they, observes the reaction of audience and proposes generation
The multiple solution of personal issue plan.
Two-way power of influence is propagated: Rui Yan et al. is proposed to be propagated by power of influence and smooths language model, to reach to solve
The certainly purpose of weak link in community network.They establish a kind of two-way socializing factor graph model, and this model utilizes document
To and document socialization behind strengthen network textual association therebetween.Such as, customer relationship and social interaction.These because of
Element communicates the attribute in author and dependence as document and is modeled with them.Finally, they are based on the power of influence estimated
Propagate lexical item to the document after smoothing.
Local in social networks is calibrated altogether: current, and people generally participate in multiple online social networking and share simultaneously
Multiple social networks richnesses service.In addition to common user, social networks is provided that similar service.These services can be total to
Enjoy the information entity of other kind a lot.Such as, position, video and product.But, these are shared in different social networkies
Entity does not mostly have any of corresponding relation and the most isolated.Jiawei Zhang et al. is for across a network chain while of these
Connect the potential corresponding relation of multiple shared entity, and in form, such issues that be referred to as " the local calibration altogether " of network
(PCT) problem.It is the prerequisite of a lot of concrete across a network application, and such as community network merges, multi information exchange and transmission.
Meanwhile, local altogether calibration problem also due to underlying cause and be considered as a kind of challenge.Including: 1, social networks is different
Matter;2, the shortage of the training example needed for modeling;3, the man-to-man restriction of communication connection aspect.In order to solve these challenges,
They propose a kind of new network calibration framework UNI-COAT (non-supervisory type is calibrated altogether).Based on heterogeneous information, this framework will
Local calibration problem altogether is converted into a combined optimization problem.In order to solve this object function, on correspondence one to one
Restriction is released, and redundancy non-existence relevant connection will be reduced by a kind of new network matching algorithm altogether.
The study of many social networkies and application: human lives is in the social networks epoch, and the global mankind are by multiple social network
Network connects and organizes.For different social networkies, this viewpoint may be according to the difference of they provided services
And difference.Mutually greet between people and describe certain specific user the most all sidedly.Relative to single
The deficient knowledge of source transmission, the appropriate fusion of many social networkies is supplied to our more preferable opportunity and carries out profound use
Family understands.But, challenge and opportunity depositing.First challenge is precisely due to some users show active in certain social networks
And show inactive in other planned networks and cause the existence of lost block data.Second challenge is how to work in coordination with whole
Close multiple social networks.For reaching this purpose, Xuemeng Song et al., by the seamless exploration from multi-source knowledge, carries
Go out a kind of new model realized for loss of data.Then, they develop many social networkies study mould of a kind of robustness
Type.
User mobility modeling for coming up: along with smart mobile phone and the surge of social networking service, based on position
The social networks put is counted as a kind of for business improving product with the instrument of service day by day.Wen-Yuan Zhu et al. have studied
Auxiliary commerce promotes the guardian technique of status, and these technology can be beaten by potential location-based social networks extensively advisably
Accuse.In order to maximize the interests come up, they are standardized as one based on the power of influence in the social networks of position it
Change greatly problem.Such as, a given target location and a location-based social networks, choose which user and use as initial
Family can be only achieved the user making their successful spread most with attraction to access the purpose of target location.Already present research is
Propose various method to calculate the probability of information transmission.It is to say, in the configuration of a static social networks, a use
How family may be gone to affect another user.But, come from the probability of spreading in location-based social networks and bring more
Challenge.Because target location and user mobility are dynamic and inquiry relies on, therefore probability of spreading is tight by both
Ghost image rings.Wen-Yuan Zhu et al. proposes two kinds of User mobility models, be respectively designated as based on Gauss and based on away from
From mobility model, for capturing single behavior of registering based on position social network user.And based on this, location aware
Probability of spreading can be acquired respectively.
Online social data is excavated: Hong-Han Shuai et al. proposes one and is called the detection of social networks mental disorder
(SNMDD) machine learning framework accurately identifies social networks mental disorder for extracting feature from social network data
Potential case.They also utilize multi-source to learn in social networks mental disorder detects and propose a kind of based on tensor model
Novel social networks detection model is used for improving performance.
The geographical social networks of interconnection: metropolis flocks together different individualities, exchanges creation for culture and knowledge
Opportunity, these finally can bring society and economic prosperity.Desislava Hristova et al. is in people and geographical interconnection
Essential aspect propose a kind of novel network perspective, this visual angle makes to be captured by social networks and Move Mode to access
The social multiformity of the city position of person becomes possibility.They define with social manager role, entropy, visitor homogeneity with
And the geographical social multiformity criterions of relevant four such as their the various accidental persons of meeting by chance that can cause.This makes it possible to assemble
The place of the place of stranger and gathering friend makes a distinction.Same it can also be used to distinguish assemble various different people place and
Assemble the place of regular guest.These attributes and the health indicator of London area are associated by they, in band height entropy and commission
Poverty-stricken area finds squireization signal.By to these area census of the populations in more than 5 years and according to comprehensive poor Britain of data
Index shows, these places have a large amount of rich and various visitor to pour in indicate their ranking and have overall lifting.
Desislava Hristova et al. discloses the relation between people and region attribute, and distinguishes different classification and important city
Geography is with reply urban policy and the development of future generation of social perception based on location application.
Suspect in social networks follows the trail of search: by specifying one specific people of name removal search at such as Facebook
Social networking service in be a basic function.But many times, it is desirable to look for a people but she is but to search target
Social networks label knows little about it (such as interest, technical ability, local, school, occupation etc.).Assuming that one social activity of each user-association
Tag set, they propose new search model (suspect follows the trail of search) in a kind of online social networks, it is intended to find a series of
Expectation target based on user's given query tag set, these labels are used for describing target.They devise a kind of greed and open
Hairdo graph search algorithm, in order to find search target.These targets not only cover inquiry tag, and can process more excellent social activity
Peer to peer interaction or for a user, has the connection of higher social nearness.
Corporate client identification: current, existing online sales may there be again offline sales in a hyundai electronics commercial company
Department.Usually, online sales attempt selling a small amount of commodity to client by the way of mass-sending bulk electronic mail or promotion code,
Thus depend critically upon the backstage algorithm of design.On the other hand, offline sales attempts the contact by being initiated by representative of sales & marketing
People sells shiploads of merchandise to corporate client.And offline sales is compared with online sales, cost is some higher.Exist with being concerned only with support
Unlike the research work such as the machine learning algorithm that line is sold, Qingbo Hu et al. describes one and utilizes heterogeneous social network
Network improves the method for offline sales effectiveness.In particular, they propose a kind of two-phase framework.Wherein, Hetero-
First Sales gains enlightenment from semantics based on unit's path learning, has constructed the figure of " company-company ", has cried again
Company's homogeneity figure (CHG).Then, use the promising company of label propagation forecast on figure, and these companies are able to ensure that
Successfully terminate the mode of offline sales.
First path based on the link prediction of Multi net voting colony: in daily life, online social networks is owing to providing various clothes
It is engaged in and becomes ubiquitous.Meanwhile, active user relates to multiple online social networks the most simultaneously and enjoys heterogeneous networks offer
Special services.Usually, social networks is generally shared some co-user and is referred to as part coherent network.J.Zhang
Et al. want in the social networks that some is consistent the formation of the link of prediction social activity simultaneously.This problem is many by formal definition
Network linking prediction (formation) problem.In the social networks that some is consistent, user can be formed by being connected with each other extensively
Wealthy link.
In order to link sorts different between these users, they propose 7 kinds " social unit paths in net " and 4 class " nets
Between unit path ".These " social unit paths " cover diversified link information in a network, some in these link informations
To solving, Multi net voting link forecasting problem is helpful, and other then can not.In order to utilize useful link information, great majority have
It is selected by a subset in " social unit path ", and process of choosing is defined as " social unit Path selection " in form.
J.Zhang et al. propose one be called " Multi net voting link identifiers " (Multi-Network Link Identifier,
MLI) effective framework, for the problem solving the prediction of general link information.Based on optional in the consistent social networks of some
Selecting " social networks unit path " and extract and set up isomery topological characteristic, MLI can assist refinement in globally consistent network and eliminate
The result of prediction mutually.
In ecommerce, binding is recommended: in current electricity business, it is recommended that system has become as an important ingredient.When
Front research in terms of commending system is focused mainly on dependency and the rate of return (RMT) improving single Recommendations.But it is true that use
What family generally contacted is a commodity set, and they may buy multiple commodity in an order.Therefore, the phase of particular commodity
Closing property and profitability may substantially rely on other commodity in set.In other words, it is recommended that set is necessary between commodity
Bundle sale.T.Zhu et al. introduces a kind of new problem being referred to as bundling recommendation problem.The such issues that of in order to solve, they
Find out the optimization commodity binding relevant to first-selected operations objective to recommend.But, binding recommendation problem be one extensive
Np hard problem.They think that the binding recommendation problem of the attribute little version of solution relying on input data is more than sufficient.
Urban poverty degree is measured: the misery index accurately and measuring urban society's economy timely has become as generation
The top-priority item of various places, boundary government.Because witness large-scale city process causing highly imbalance and these all
Need to be mediated.Traditionally, misery index is obtained by census data, but this acquisition mode cost is relatively
Height, and be that every few years could obtain.In recent years, alternative computational methods are suggested in certain space-time granularity
It is used for automatically extracting misery index.But, they usually require that access data set (the most detailed record), and these are not
Can open get from government and agency.In order to make up, Desislava Hristova et al. proposes a kind of new side
Method is in order to automatic mining misery index in a preferable space-time granularity, and this method has only to freely obtain user-generated content
?.What need to be carried more precisely, this method needs to access data set to be described in physical world city element
Out.
Topological attribute in urban environment and temporal dynamic property: understand spatial network be formed by the track of mobile subscriber right
Application in epiphytotics Local Search is helpful to.Although have potential impact in a lot of fields, but due at one
Preferably lacking large-scale data in space-time solution, therefore some aspects of mankind's two mobility network are not the most by big model
Enclose and probe into.Anastasios Noulas et al. has carried out empirical analysis to the topological attribute of LAN, in heavy-tailed degree distribution,
The aspects such as ternary Guan Bi mechanism and worldlet attribute record the similarity of they and social networks.But it is different from social networks
, in terms of same joining property, LAN shows the trend of a kind of Hybrid connections.And this has surprising with those networks
Similitude.Anastasios Noulas et al. utilizes extra semantic information to explain that those undertake function angle in a network
The behavior of the node (such as, tour center or food point) of color.Finally, in LAN as time goes by, advise greatly
The appearance of mould new url, they propose a kind of new Gravity Models, in order to three that interzone in urban environment is connective
Critical elements (Internet mutual, the ambulant dynamic of the mankind, geographic distance) flocks together.
Multilamellar manager in geographical social networks: open network structure and manager position are considered as to maintain society for a long time
Meeting capital and competitive advantage aspect have played vital effect.Originally disconnecting individual intermediary's degree can be online and off-line
Social networks distinguishes.Such as, user is probably the intermediary of two online users, and the two user is carried out by social media
Information retrieval exchanges information off-line.But, the network research of social capital on line with usual quilt in influencing each other under line
Ignore, and be concentrated mainly on monolayer.Desislava Hristova et al. proposes the multilevel method of a kind of geographical society and uses
With intermediary, the truth on this basis, online and offline allowing for integrating society capital is disclosed for out.They are online by verifying
In social networks, position and the user of user extend the general of intermediary by logging on the off-line Move Mode of check-ins data
Read.They find, obvious and asymmetric by social and the social activity of colocated network and map intermediary position.One side
Face, if in fact the off-line position of user is also included into limit of consideration, when user intermediary ability obtains relaxing when,
They may show as the feature of broker.On the other hand, in a network user show deficiency off-line brokerage pragmatic
Power may activate broker, and these are online and off-line is connected together alternately.
Urban economy growth is followed the tracks of: urban resource is allocated according to socio-economic indicator, development from online concern
The Fast Urbanization of Chinese Home needs these indexs that upgrade in time.Census data collect sky high cost make this in time
Renewal becomes extremely difficult.In order to avoid according to out-of-date Distribution Indexes resource, using data that these indexs are carried out part
Update and supplement.It is possible for using social media to reach this purpose in developed country (mainly Great Britain and America).Carmen
Vaca Ruiz et al. analyzes a random sample in microblogging service and Accurate Prediction has gone out the GDP value in city.In order to carry out
Prediction, they utilize global endemic sociology conceptual illustration, local and full while that economically successfully city trending towards
Relate to mutual in the range of ball.It is true that utilize the local performance in the whole world in one city of social media data metric effectively to represent this
The happiness in individual city.
Summary of the invention
In view of this, high-quality node detecting method and system during the present invention proposes a kind of social networks.
High-quality node detecting method in a kind of social networks, it comprises the steps:
The social networks node set at the high-quality node place that S1, extraction needs detect;
S2, the social networks node in social networks node set is set up the node mapping relations of social networks;
S3, extract detection high-quality node diagnostic according to the mechanics of high-quality node;According to the high-quality node diagnostic extracted
Set up the characterization rules model of high-quality node detection;
S4, social networks node is grouped as experiment sample, then carries out station work and node-classification;
S5, the result detecting high-quality node are estimated and feedback result, and will not be inconsistent during training repeatedly
The rule closing detection high-quality node diagnostic is corrected, thus reaches the purpose being optimized model;
S6, will optimize after model turn again to step S4 node training with classify link carry out high-quality node detection to carry
High detection progress, and it is iterated computing until the detection progress of high-quality node exceedes setting threshold value thus completes whole social network
High-quality node detection process in network.
In social networks of the present invention in high-quality node detecting method,
In described step S3, the mechanics extraction detection high-quality node diagnostic according to high-quality node includes:
Node is set and is judged as the nodal community of outstanding node, possess the nodal community of outstanding node at a certain node
Time, then this nodal community has the feature of high quality;
Arrange and between node, the most just should possess interactivity frequently, referred to as interactive degree Vinter;In society
Hand in network, be connected limit as node has between the feature of core node, and they with all mid-side nodes to exist;Limit is connected at these
In, regard the active of high-quality node as out-degree V alternatelyout, out-degree is self to point to the limit of other node, and passively sees alternately
Make in-degree Vin, in-degree is the limit that other node points to self, then high-quality node exists more than the first predetermined threshold value the most simultaneously
In-degree and out-degree, and out-degree in-degree ratioClose to 1;
In-degree and the out-degree exceeding preset value, and out-degree in-degree ratio will be there is simultaneouslyThreshold is preset more than 1 and more than second
The node of value is as non-prime node.
In social networks of the present invention in high-quality node detecting method, in described step S3, high-quality node detected
The matrix of journey is expressed as follows:
Mapping matrix is the mapping relations matrix between in-degree and the out-degree of inspected object;Wherein, Min×outRepresenting matrix
Title, In represents the in-degree set of node, and Out represents the out-degree set of node;Vi,in, i=1...n and Vj,out, j=1...m
Represent in-degree and the out-degree of node j of node i respectively;I=1...n, j=1...m represent node i in-degree and joint
Point j out-degree ratio;As i=j, high-quality node and non-prime node can be detected, the now P of high-quality nodeijTrend towards
1, and the P of non-prime nodeijMore than 1 and more than the second predetermined threshold value or less than 1 and less than the 3rd predetermined threshold value;And as i ≠ j
Time, if Pij=0 shows to there is not interactive relation between different node, otherwise exists for interactive relation between node.
In social networks of the present invention in high-quality node detecting method, described step S5 also includes formulate detection
The recall rate of result and accuracy rate, and accuracy rate is carried out threshold value setting, to decide whether that jumping to step S6 is iterated fortune
Calculation process.
The present invention also provides for high-quality node detection system in a kind of social networks, and it includes such as lower unit:
Node set acquiring unit, for extracting the social networks node set at the high-quality node place needing detection;
Mapping relations set up unit, for the social networks node in social networks node set is set up social networks
Node mapping relations;
High-quality node probe unit, extracts detection high-quality node diagnostic for the mechanics according to high-quality node;According to
The high-quality node diagnostic extracted sets up the characterization rules model of high-quality node detection;
Station work unit, for being grouped as experiment sample by social networks node, then carries out station work
And node-classification;
Assessment feedback unit, for being estimated the result of high-quality node detection feedback result, and is instructing repeatedly
During white silk, the rule not meeting detection high-quality node diagnostic is corrected, thus reaches the purpose that model is optimized;
Iteration unit, the model after optimizing turns again to the node training of station work unit and enters with classification link
Row high-quality node detects to improve detection progress, and is iterated computing until the detection progress of high-quality node exceedes setting threshold value
Thus complete high-quality node detection process in whole social networks.
In social networks of the present invention in high-quality node detection system,
In described high-quality node probe unit, the mechanics extraction detection high-quality node diagnostic according to high-quality node includes:
Node is set and is judged as the nodal community of outstanding node, possess the nodal community of outstanding node at a certain node
Time, then this nodal community has the feature of high quality;
Arrange and between node, the most just should possess interactivity frequently, referred to as interactive degree Vinter;In society
Hand in network, be connected limit as node has between the feature of core node, and they with all mid-side nodes to exist;Limit is connected at these
In, regard the active of high-quality node as out-degree V alternatelyout, out-degree is self to point to the limit of other node, and passively sees alternately
Make in-degree Vin, in-degree is the limit that other node points to self, then high-quality node exists more than the first predetermined threshold value the most simultaneously
In-degree and out-degree, and out-degree in-degree ratioClose to 1;
In-degree and the out-degree exceeding preset value, and out-degree in-degree ratio will be there is simultaneouslyThreshold is preset more than 1 and more than second
The node of value is as non-prime node.
In social networks of the present invention in high-quality node detection system, high-quality in described high-quality node probe unit
The matrix of node detection process is expressed as follows:
Mapping matrix is the mapping relations matrix between in-degree and the out-degree of inspected object;Wherein, Min×outRepresenting matrix
Title, In represents the in-degree set of node, and Out represents the out-degree set of node;Vi,in, i=1...n and Vj,out, j=1...m
Represent in-degree and the out-degree of node j of node i respectively;I=1...n, j=1...m represent node i in-degree and joint
Point j out-degree ratio;As i=j, high-quality node and non-prime node can be detected, the now P of high-quality nodeijTrend towards
1, and the P of non-prime nodeijMore than 1 and more than the second predetermined threshold value or less than 1 and less than the 3rd predetermined threshold value;And as i ≠ j
Time, if Pij=0 shows to there is not interactive relation between different node, otherwise exists for interactive relation between node.
In social networks of the present invention in high-quality node detection system, described assessment feedback unit also includes system
Determine recall rate and the accuracy rate of result of detection, and accuracy rate is carried out threshold value setting, to decide whether that jumping to iteration unit enters
Row iteration calculating process.
Implement high-quality node detecting method and system in the social networks that the present invention provides compared with prior art have with
Lower beneficial effect: by based on the knowledge and experience that high-quality be the user discover that by observer, therefrom extract high-quality user and had
Standby feature sets up correlation rule and feature identification storehouse.The problem that observer found that high-quality user is incorporated into grinding of social networks
Study carefully in field, excavate as theoretical foundation with the figure of social networks, this problem is converted in social networks and finds and detect high-quality
It is also solved by the problem of node, it is possible to for finding the core with user's request in social networks with higher compatible degree
Node, it is possible to contribute to the individual value fully realizing oneself in social relation network.
Accompanying drawing explanation
Fig. 1 be the embodiment of the present invention social networks in high-quality node detecting method schematic diagram;
Fig. 2 is the high-quality node detection model schematic diagram under the embodiment of the present invention " teacher-student " social networks;
Fig. 3 is the matrix description figure of the high-quality node detection of the embodiment of the present invention;
Fig. 4 be the embodiment of the present invention social networks in high-quality node detection system structured flowchart.
Detailed description of the invention
As shown in Figures 1 to 4, the detection method of high-quality node and be in a kind of social networks that the embodiment of the present invention proposes
System, for finding the core node with user's request in social networks with higher compatible degree.In physical world, this detection
Method has broad application prospects.Such as, in " teacher-student " this kind of social networks of relation, expect to seek as teacher
Outstanding student does emphasis and cultivates, and wishes to find outstanding teacher oneself to learn to teach as student;" enterprise-visitor
Family " in this kind of social networks of relation, wish that as enterprise the client finding high-quality forms firm strategic partner with oneself
Relation, and the production marketing of oneself to client therefrom to obtain profit, and as expect select high-quality enterprise buy product
Product are to meet the consumption demand of oneself;In " enterprise-talent " this kind of social networks of relation, wish to solicit high-quality as enterprise
The high-caliber talent to create more commercial value for company, and needs to find oneself enterprise applicable to realize as the talent
The value of life of oneself and aspiration.Social network relationships like this can be found everywhere, it can be seen that, at numerous social networkies
High-quality node is detected by node there is very important social value and realistic meaning.In order to solve in social networks
The problem carrying out for high-quality node detecting, the existing thinking by the problem of solution and step are as shown in Figure 1.From figure 1 it appears that
First have to extract the social networks node set at the high-quality node place needing detection.The most mentioned above, outstanding teacher or
The social networks node set of Ontario Scholar place " teacher-student " relation.Extract social networks node set complete or collected works it
After, need social networks node is set up the node mapping relations of social networks.Then, carry according to the mechanics of high-quality node
Take detection high-quality node needs to meet which feature.Such as, outstanding student often likes asking a question.So this phenomenon just may be used
Become the feature of detection high-quality node.Treat that high-quality node diagnostic just can set up the feature of high-quality node detection after being extracted
Rule model.Afterwards, social networks node can be regarded as experiment sample and be grouped, then carry out station work and node divides
Class.Afterwards, the result of high-quality node detection can be estimated and feedback result, and will less accord with during training repeatedly
The rule closing detection high-quality node diagnostic is corrected, to reach the purpose being optimized model.Finally, the mould after optimizing
Type turns again to node training and carries out high-quality node detection to improve detection progress with classification link, is so iterated computing straight
Detection progress to high-quality node exceedes setting threshold value to complete whole detection process.
It follows that by combine concrete should be for describing the process of high-quality node detection in detail.The process of whole description will
It is related to that this social networks carries out process prescription with well-known " teacher-student ".High-quality node as shown in Figure 1 detected
Cheng Zhong, most crucial link is that the extraction of high-quality node diagnostic rule sets up this link with detection model.Detection high-quality node
The problem that this link shows as detecting outstanding teacher or Ontario Scholar in " teacher-student " this social networks.
It follows that carry out the detailed of node detection feature Rule Extraction process by find the process of Ontario Scholar by teacher as a example by
Thin description.
Can be drawn by experience in actual life, outstanding student typically exhibits self outstanding personal attribute.Example
As, they like to ponder a problem, and hanker after thinking independently and out being consulted to teacher by insoluble problem induction and conclusion, and always
Teacher forms frequent interaction.Outstanding student in addition to showing outstanding self attributes, they would generally with surrounding other with
Talk about problem thus form communication widely.Therefore, these high-qualitys intuitively can be extracted through these daily observations
Node diagnostic rule.And these rules directly perceived can be summarized as follows in the theory of social networks:
1, Ontario Scholar self has outstanding personal attribute's (like thinking, like to put question to, be good at induction and conclusion etc.), then joint
Point attribute has the feature of high quality;
2, Ontario Scholar frequently puts question to teacher's node, carries out interaction with teacher, then high-quality node and Teacher's Day
Between point (periphery student's node), the most just should possess interactivity frequently, referred to as interactive degree Vinter.Embody
In social networks, these nodes are likely to be of between the feature of core node, and they with all mid-side nodes to exist and are connected limit.At this
In a little connection limits, if the active of high-quality node is regarded as out-degree V alternatelyout(self pointing to the limit of other node), and passively hand over
Regard in-degree V mutually asin(limit of other node sensing self), then high-quality node exists bigger in-degree and out-degree the most simultaneously,
And(out-degree in-degree ratio) is possibly close to 1;
3, there is the most mutual node and be likely to be non-prime node.Such as, the student of a low academic is often
There is also the most mutual between lifting school grade, and week mid-side node and teacher node.Therefore in the classification of detection high-quality node
During, tend in this category node misclassification to high-quality node set.But, although this category node has bigger going out
Degree, but their in-degree is not very big, thereforeMay be much larger than 1.Thus, high-quality node can be entered with non-prime node
Row is distinguished more accurately.
Above-mentioned rule uses the mode of social networks graph model can carry out directviewing description, as shown in Figure 2.From Fig. 2 permissible
Find out intuitively, high-quality node and teacher's node set and the mutual situation of student's intersection of sets.By combining three of foregoing description
High-quality node can significantly be distinguished by rule with non-prime node.
Extraction principle for the high-quality node diagnostic rule of the social networks under other relation can be retouched in accordance with in this patent
The extraction process analogy stated completes.Such as, the behavior of " enterprise-client " relation next one top-tier customer often and is deposited between enterprise
In strong interactivity, then by observing and combine the high-quality attribute meeting enterprise demand that client self exists, enterprise can be made equally
Reach to find the purpose of top-tier customer.
Therefore, the schematic diagram in Fig. 2 also can use the method for matrix description to carry out high-quality node detection in general sense
The formalized description of process.As it is shown on figure 3, be the matrix description method of high-quality node detection process.
From as can be seen in Figure 3, this mapping matrix is the mapping relations square between in-degree and the out-degree of inspected object
Battle array.Wherein, Min×outRepresenting matrix title, In represents the in-degree set of node, and Out represents the out-degree set of node.Vi,in, i=
And V 1...nj,out, j=1...m represents in-degree and the out-degree of node j of node i respectively.I=1...n, j=
1...m node i in-degree and node j out-degree ratio are represented.Rule mentioned above is had to understand, as i=j, can be to high-quality node
Detect with non-prime node, now the P of high-quality nodeijGenerally tend to 1, and the P of non-prime nodeijTypically much deeper than 1
Or much smaller than 1.And as i ≠ j, if Pij=0 shows to there is not interactive relation between different node, otherwise exists for mutual between node
Relation.Therefore, the purpose of high-quality node detection is also can reach by matrix operations.
Assessment result and feedback: whether the evaluation criteria for high-quality node often meets detection demand in accordance with result of detection
Formulate.Such as, the result of detection of the Ontario Scholar under " teacher-student " relation, if the detection that detection model obtains
Result is not an Ontario Scholar, then the formulation of evaluation criteria just should be modified.Such as, result of detection can be formulated
Recall rate and accuracy rate, and accuracy rate is carried out threshold value setting, to decide whether to the interative computation mistake continued as shown in Figure 1
Journey.
As shown in Figure 4, the embodiment of the present invention also provides for high-quality node detection system in a kind of social networks, and it includes as follows
Unit:
Node set acquiring unit, for extracting the social networks node set at the high-quality node place needing detection.
Mapping relations set up unit, for the social networks node in social networks node set is set up social networks
Node mapping relations.
High-quality node probe unit, extracts detection high-quality node diagnostic for the mechanics according to high-quality node;According to
The high-quality node diagnostic extracted sets up the characterization rules model of high-quality node detection.
Station work unit, for being grouped as experiment sample by social networks node, then carries out station work
And node-classification.
Assessment feedback unit, for being estimated the result of high-quality node detection feedback result, and is instructing repeatedly
During white silk, the rule not meeting detection high-quality node diagnostic is corrected, thus reaches the purpose that model is optimized.
Iteration unit, the model after optimizing turns again to the node training of station work unit and enters with classification link
Row high-quality node detects to improve detection progress, and is iterated computing until the detection progress of high-quality node exceedes setting threshold value
Thus complete high-quality node detection process in whole social networks.
In social networks of the present invention in high-quality node detection system,
In described high-quality node probe unit, the mechanics extraction detection high-quality node diagnostic according to high-quality node includes:
Node is set and is judged as the nodal community of outstanding node, possess the nodal community of outstanding node at a certain node
Time, then this nodal community has the feature of high quality;
Arrange and between node, the most just should possess interactivity frequently, referred to as interactive degree Vinter;In society
Hand in network, be connected limit as node has between the feature of core node, and they with all mid-side nodes to exist;Limit is connected at these
In, regard the active of high-quality node as out-degree V alternatelyout, out-degree is self to point to the limit of other node, and passively sees alternately
Make in-degree Vin, in-degree is the limit that other node points to self, then high-quality node exists more than the first predetermined threshold value the most simultaneously
In-degree and out-degree, and out-degree in-degree ratioClose to 1;
In-degree and the out-degree exceeding preset value, and out-degree in-degree ratio will be there is simultaneouslyThreshold is preset more than 1 and more than second
The node of value is as non-prime node.
In social networks of the present invention in high-quality node detection system, high-quality in described high-quality node probe unit
The matrix of node detection process is expressed as follows:
Mapping matrix is the mapping relations matrix between in-degree and the out-degree of inspected object;Wherein, Min×outRepresenting matrix
Title, In represents the in-degree set of node, and Out represents the out-degree set of node;Vi,in, i=1...n and Vj,out, j=1...m
Represent in-degree and the out-degree of node j of node i respectively;I=1...n, j=1...m represent node i in-degree and joint
Point j out-degree ratio;As i=j, high-quality node and non-prime node can be detected, the now P of high-quality nodeijTrend towards
1, and the P of non-prime nodeijMore than 1 and more than the second predetermined threshold value or less than 1 and less than the 3rd predetermined threshold value;And as i ≠ j
Time, if Pij=0 shows to there is not interactive relation between different node, otherwise exists for interactive relation between node.
Second predetermined threshold value is much larger than 1, and can independently arrange;3rd predetermined threshold value is much smaller than 1, and can independently set
Put.
In social networks of the present invention in high-quality node detection system, described assessment feedback unit also includes system
Determine recall rate and the accuracy rate of result of detection, and accuracy rate is carried out threshold value setting, to decide whether that jumping to iteration unit enters
Row iteration calculating process.
It is understood that for the person of ordinary skill of the art, can conceive according to the technology of the present invention and do
Go out other various corresponding changes and deformation, and all these change all should belong to the protection model of the claims in the present invention with deformation
Enclose.
Claims (8)
1. high-quality node detecting method in a social networks, it is characterised in that it comprises the steps:
The social networks node set at the high-quality node place that S1, extraction needs detect;
S2, the social networks node in social networks node set is set up the node mapping relations of social networks;
S3, extract detection high-quality node diagnostic according to the mechanics of high-quality node;Set up according to the high-quality node diagnostic extracted
The characterization rules model of high-quality node detection;
S4, social networks node is grouped as experiment sample, then carries out station work and node-classification;
S5, the result detecting high-quality node are estimated and feedback result, and will not meet spy during training repeatedly
The rule surveying high-quality node diagnostic is corrected, thus reaches the purpose being optimized model;
S6, will optimize after model turn again to step S4 node training with classification link carry out high-quality node detection to improve spy
Survey progress, and it is iterated computing until the detection progress of high-quality node exceedes setting threshold value thus completes in whole social networks
High-quality node detection process.
2. high-quality node detecting method in social networks as claimed in claim 1, it is characterised in that
In described step S3, the mechanics extraction detection high-quality node diagnostic according to high-quality node includes:
Node is set and is judged as the nodal community of outstanding node, when a certain node possesses the nodal community of outstanding node, then
This nodal community has the feature of high quality;
Arrange and between node, the most just should possess interactivity frequently, referred to as interactive degree Vinter;In social network
In network, it is connected limit as node has between the feature of core node, and they with all mid-side nodes to exist;In these connect limit, will
Out-degree V is regarded in the active of high-quality node alternately asout, out-degree is self to point to the limit of other node, and passively regards in-degree alternately as
Vin, in-degree is that other node points to self limit, then high-quality node exists the most simultaneously more than the first predetermined threshold value in-degree with
Out-degree, and out-degree in-degree ratioClose to 1;
In-degree and the out-degree exceeding preset value, and out-degree in-degree ratio will be there is simultaneouslyMore than 1 and more than the second predetermined threshold value
Node is as non-prime node.
3. high-quality node detecting method in social networks as claimed in claim 2, it is characterised in that high-quality in described step S3
The matrix of node detection process is expressed as follows:
Mapping matrix is the mapping relations matrix between in-degree and the out-degree of inspected object;Wherein, Min×outRepresenting matrix name
Claiming, In represents the in-degree set of node, and Out represents the out-degree set of node;Vi,in, i=1 ... n and Vj,out, j=1 ... m is respectively
Represent in-degree and the out-degree of node j of node i;I=1 ... n, j=1 ... m represents node i in-degree and node j out-degree
Ratio;As i=j, high-quality node and non-prime node can be detected, the now P of high-quality nodeijTrend towards 1, and non-optimum
The P of matter nodeijMore than 1 and more than the second predetermined threshold value or less than 1 and less than the 3rd predetermined threshold value;And as i ≠ j, if Pij=
0 shows to there is not interactive relation between different node, otherwise exists for interactive relation between node.
4. high-quality node detecting method in social networks as claimed in claim 3, it is characterised in that also wrap in described step S5
Include recall rate and the accuracy rate formulating result of detection, and accuracy rate is carried out threshold value setting, to decide whether to jump to step S6
It is iterated calculating process.
5. high-quality node detection system in a social networks, it is characterised in that it includes such as lower unit:
Node set acquiring unit, for extracting the social networks node set at the high-quality node place needing detection;
Mapping relations set up unit, for the social networks node in social networks node set is set up the node of social networks
Mapping relations;
High-quality node probe unit, extracts detection high-quality node diagnostic for the mechanics according to high-quality node;According to extraction
High-quality node diagnostic set up high-quality node detection characterization rules model;
Station work unit, for being grouped as experiment sample by social networks node, then carries out station work and joint
Point classification;
Assessment feedback unit, for being estimated the result of high-quality node detection feedback result, and was training repeatedly
The rule not meeting detection high-quality node diagnostic is corrected by journey, thus reaches the purpose that model is optimized;
Iteration unit, the model after optimizing turns again to the node training of station work unit and carries out excellent with classification link
Matter node detects to improve detection progress, and be iterated computing until the detection progress of high-quality node exceed set threshold value thus
Complete high-quality node detection process in whole social networks.
6. high-quality node detection system in social networks as claimed in claim 5, it is characterised in that
In described high-quality node probe unit, the mechanics extraction detection high-quality node diagnostic according to high-quality node includes:
Node is set and is judged as the nodal community of outstanding node, when a certain node possesses the nodal community of outstanding node, then
This nodal community has the feature of high quality;
Arrange and between node, the most just should possess interactivity frequently, referred to as interactive degree Vinter;In social network
In network, it is connected limit as node has between the feature of core node, and they with all mid-side nodes to exist;In these connect limit, will
Out-degree V is regarded in the active of high-quality node alternately asout, out-degree is self to point to the limit of other node, and passively regards in-degree alternately as
Vin, in-degree is that other node points to self limit, then high-quality node exists the most simultaneously more than the first predetermined threshold value in-degree with
Out-degree, and out-degree in-degree ratioClose to 1;
In-degree and the out-degree exceeding preset value, and out-degree in-degree ratio will be there is simultaneouslyMore than 1 and more than the second predetermined threshold value
Node is as non-prime node.
7. high-quality node detection system in social networks as claimed in claim 6, it is characterised in that described high-quality node detects
In unit, the matrix of high-quality node detection process is expressed as follows:
Mapping matrix is the mapping relations matrix between in-degree and the out-degree of inspected object;Wherein, Min×outRepresenting matrix name
Claiming, In represents the in-degree set of node, and Out represents the out-degree set of node;Vi,in, i=1 ... n and Vj,out, j=1 ... m is respectively
Represent in-degree and the out-degree of node j of node i;I=1 ... n, j=1 ... m represents node i in-degree and node j out-degree
Ratio;As i=j, high-quality node and non-prime node can be detected, the now P of high-quality nodeijTrend towards 1, and non-optimum
The P of matter nodeijMore than 1 and more than the second predetermined threshold value or less than 1 and less than the 3rd predetermined threshold value;And as i ≠ j, if Pij=
0 shows to there is not interactive relation between different node, otherwise exists for interactive relation between node.
8. high-quality node detection system in social networks as claimed in claim 7, it is characterised in that described assessment feedback unit
In also include formulating the recall rate of result of detection and accuracy rate, and accuracy rate is carried out threshold value setting, to decide whether to jump to
Iteration unit is iterated calculating process.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610550702.6A CN106126746A (en) | 2016-07-14 | 2016-07-14 | High-quality node detecting method and system in a kind of social networks |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610550702.6A CN106126746A (en) | 2016-07-14 | 2016-07-14 | High-quality node detecting method and system in a kind of social networks |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106126746A true CN106126746A (en) | 2016-11-16 |
Family
ID=57283656
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610550702.6A Pending CN106126746A (en) | 2016-07-14 | 2016-07-14 | High-quality node detecting method and system in a kind of social networks |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106126746A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106850599A (en) * | 2017-01-18 | 2017-06-13 | 中国科学院信息工程研究所 | A kind of NAT detection methods based on fusion user behavior and sudden peal of thunder ID |
CN107248054A (en) * | 2017-06-02 | 2017-10-13 | 河北斯博思创新科技有限公司 | A kind of complex relationship network data visual analysis method based on the malicious industry of easy system |
CN109670635A (en) * | 2018-12-05 | 2019-04-23 | 厦门笨鸟电子商务有限公司 | A kind of generation method of the optimal issuing time of social information |
CN110083777A (en) * | 2018-01-26 | 2019-08-02 | 腾讯科技(深圳)有限公司 | A kind of social network user group technology, device and server |
CN110096650A (en) * | 2019-04-23 | 2019-08-06 | 北京科技大学 | The analysis method and device of network connection intensity |
CN110134877A (en) * | 2019-05-15 | 2019-08-16 | 天津大学 | Move down the line the method and apparatus that seed user is excavated in social networks |
CN113868544A (en) * | 2021-12-03 | 2021-12-31 | 杭银消费金融股份有限公司 | Intelligent service file processing method and service server |
-
2016
- 2016-07-14 CN CN201610550702.6A patent/CN106126746A/en active Pending
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106850599A (en) * | 2017-01-18 | 2017-06-13 | 中国科学院信息工程研究所 | A kind of NAT detection methods based on fusion user behavior and sudden peal of thunder ID |
CN106850599B (en) * | 2017-01-18 | 2019-12-03 | 中国科学院信息工程研究所 | A kind of NAT detection method based on fusion user behavior and sudden peal of thunder ID |
CN107248054A (en) * | 2017-06-02 | 2017-10-13 | 河北斯博思创新科技有限公司 | A kind of complex relationship network data visual analysis method based on the malicious industry of easy system |
CN110083777A (en) * | 2018-01-26 | 2019-08-02 | 腾讯科技(深圳)有限公司 | A kind of social network user group technology, device and server |
CN110083777B (en) * | 2018-01-26 | 2022-11-25 | 腾讯科技(深圳)有限公司 | Social network user grouping method and device and server |
CN109670635A (en) * | 2018-12-05 | 2019-04-23 | 厦门笨鸟电子商务有限公司 | A kind of generation method of the optimal issuing time of social information |
CN110096650A (en) * | 2019-04-23 | 2019-08-06 | 北京科技大学 | The analysis method and device of network connection intensity |
CN110134877A (en) * | 2019-05-15 | 2019-08-16 | 天津大学 | Move down the line the method and apparatus that seed user is excavated in social networks |
CN113868544A (en) * | 2021-12-03 | 2021-12-31 | 杭银消费金融股份有限公司 | Intelligent service file processing method and service server |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Ritterbusch et al. | Defining the metaverse: A systematic literature review | |
CN106126746A (en) | High-quality node detecting method and system in a kind of social networks | |
Colomo-Palacios et al. | Towards a social and context-aware mobile recommendation system for tourism | |
Ma et al. | SuperedgeRank algorithm and its application in identifying opinion leader of online public opinion supernetwork | |
Chui et al. | The social economy: Unlocking value and productivity through social technologies | |
Belderbos et al. | Generic and specific social learning mechanisms in foreign entry location choice | |
Cool et al. | Strategic group formation and performance: The case of the US pharmaceutical industry, 1963–1982 | |
Maybury | Expert finding systems | |
Borgatti et al. | Analyzing social networks using R | |
Aaldering et al. | Analyzing the impact of industry sectors on the composition of business ecosystem: A combined approach using ARM and DEMATEL | |
Celata et al. | A room with a (re) view. Short-term rentals, digital reputation and the uneven spatiality of platform-mediated tourism | |
Kabassi | Evaluating museum websites using a combination of decision-making theories | |
Evans | Creative cities–An international perspective | |
CN107767280A (en) | A kind of high-quality node detecting method based on element of time | |
Arora et al. | Social capacitance: Leveraging absorptive capacity in the age of social media | |
Lei et al. | Modelling and analysis of big data platform group adoption behaviour based on social network analysis | |
Mladenow et al. | Collaboration and locality in crowdsourcing | |
CN110309363A (en) | A kind of instructional video segment method of commerce of knowledge based point | |
Sharma et al. | Journal of computer information systems: intellectual and conceptual structure | |
Lu et al. | A fuzzy social network centrality analysis model for interpersonal spatial relations | |
Luo et al. | An assessing framework for the proper allocation of collection and delivery points from the residents' perspective | |
Pandey et al. | Bipolar‐Valued Fuzzy Social Network and Centrality Measures | |
Esmaeili et al. | Conformance checking of the activity network with the social relationships structure in the context of social commerce | |
Sui et al. | Social media as sensor in real world: movement trajectory detection with microblog | |
Zhang | Research on collaborative filtering recommendation algorithm based on social network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161116 |
|
RJ01 | Rejection of invention patent application after publication |